NCUA’s release of the much touted PIMCO report on April 17 was little more than an ad brochure for the firm. There was no data for understanding the firm’s or NCUA’s analysis of any corporate’s portfolio. If a corporate credit union had given this information to NCUA as the basis for their estimates, examiners would have said, “Not good enough—show us the numbers!”
The discussion over PIMCO’s analysis however is obscuring a larger issue. NCUA has decreed that credit unions must expense $10 billion based on their analysis of the corporate network’s investments. Part of this expense is the systemic losses via the NCUSIF assessments; the remainder is the $4 billion capital at WesCorp and U.S. Central which the Agency says is gone.
The Use and Misuse of Models
But how was this $10 billion expense, for which every credit union is booking a pro rata share, estimated? With modeling. Below is a simple example of a bond in WesCorp’s portfolio for $105 million, CUSIP 751153AC1. The bond is a senior tranche, originally rate AAA at the time of purchase and now rated B or Ca depending on which rating agency is used.
This sample analysis uses Bloomberg analytics. Three input assumptions drive virtually all bond models:
- the rate at which the underlying mortgages are assumed to prepay faster than contractual payments (cpr),
- the default rate on the underlying collateral (% of loans going to foreclosure) and
- the average loss estimated on these foreclosed loans, or severity.
The chart below shows these three numbers: 9.5 CPR, a 10% default rate and a loss rate on loans foreclosed of 60%. These are all based on economist’s estimates of future trends.
Using contractual payments plus CPR assumptions, the cash flows are then modeled on Bloomberg. These show a potential loss on underlying collateral of $15 million beginning in the year 2016.
Citing PIMCO and their other models, NCUA is requiring credit unions to expense this possible loss today. This new regulatory accounting is forcing the entire industry to “mark-to-model” a process even NCUA says could result in different estimates whenever models are rerun.
Other Models, Other Answers
Bloomberg is not the only modeling option available. Other firms allow more complex variations in the assumptions such as changing the loss severity over time. But every model is driven by assumptions. The three key drivers are inputs above which are chosen using economic and statistical data such as unemployment trends or housing sales.
Virtually all models produce forecasts of cash flows and possible credit defaults over the life of the bond. Some even convert these to yield-to-maturity estimates so a portfolio manager might compare different predicted outcomes in deciding which bonds to hold. What models do not produce is a single “fair market price;” rather there are multiple outcomes from various scenarios or discounted cash flow assumptions.
While models offer the allure of precise numbers, they are simply tools which depend on economist’s inputs. Sometimes these economic predictions can make weather forecasters look good.
So, rather than matching investment losses as incurred, either from payment defaults or bond sales, NCUA has required all credit unions to expense today these loss estimates which go out years into the future.
The consequence of this $10 billion upfront expense is to reduce net worth ratios across the system. This in turn constrains every credit union’s lending and savings growth. Just when members need credit unions the most, NCUA’s actions are causing credit unions to pull back.
At a 10% average capital level, the impact on members is to reduce savings growth by $100 billion and loans by the average loan-to-share ratio, which today is 82% or $82 billion. If a credit union’s capital ratio is lower, the impact is even greater!
Instead of supporting credit unions unique “counter cyclical role” in today’s troubled economy, NCUA’s mark to model methodology has just the opposite impact. It puts credit unions under the same constraints as other market driven lenders.
A Better Solution
A better way to meet the expenses from corporate losses would be to assess an insurance premium as these events are incurred over the next four to six years. A 10 basis point premium on a $250,000 maximum insurance coverage, assuming 8% share growth, would generate almost $5 billion in premium in just six years.
This is the solution that fits not only credit unions’ role today, but also is consistent with how cooperatives generate capital, that is from future earnings.
NCUA’s PIMCO “report” failed to meet any standard that would clarify the numbers the Agency is using. However the real issue is the policy failure that imposes costs on the system today for estimated future losses. No other regulatory agency, bank or other institution is using this practice. So why are credit unions?